A global optimal algorithm for camera calibration with one-dimensional objects

Abstract

The emergent one-dimensional (1D) calibration is very suitable for multi-camera calibration. However its accuracy is not satisfactory. Conventional optimal algorithms, such as bundle adjustment, do not perform well for the non-convex optimization of 1D calibration. In this paper, a practical optimal algorithm for camera calibration with 1D objects using branch and bound framework is presented. To obtain the optimal solution which can provide ε-optimality, tight convex relaxations of the objective functions are constructed and minimized in a branch and bound optimization framework. Experiments prove the validity of the proposed method. © 2011 Springer-Verlag.